From: Leandro Lucarella <luca@llucax.com.ar>
Date: Sat, 14 Jul 2012 18:09:51 +0000 (+0200)
Subject: Ignore br tags for tag depth calculations
X-Git-Url: https://git.llucax.com/software/subdivxget.git/commitdiff_plain/ac39e3e3707a898402033588e822d01c8e6be9f9?ds=inline

Ignore br tags for tag depth calculations

SubDivX uses old, unclosed br tags, so they break the tag depth
calculation.
---

diff --git a/subdivxget b/subdivxget
index 395eb35..d7c32d1 100755
--- a/subdivxget
+++ b/subdivxget
@@ -67,7 +67,8 @@ class SubDivXHTMLParser(HTMLParser.HTMLParser):
 				self.attr = 'titulo'
 			elif attrs.get('href', '').startswith(self.down_uri):
 				self.cur['url'] = attrs['href']
-		if self.parsing:
+		# br are usually not closed, so ignore them in depth calculation
+		if self.parsing and tag != 'br':
 			self.depth += 1
 
 	def handle_endtag(self, tag):
@@ -75,7 +76,9 @@ class SubDivXHTMLParser(HTMLParser.HTMLParser):
 			if tag == 'script' or tag == 'style':
 				self.in_script_style = False
 				return
-			self.depth -= 1
+			# see comment in handle_starttag()
+			if tag != 'br':
+				self.depth -= 1
 		if self.depth == 0:
 			self.parsing = False